It looks like that because Unicode is trying to solve a problem that everyone thinks is easy until they uncover the true extent of encoding human languages.
Surrogate pairs were new to Unicode 2.0. Unicode 1.0 didn't anticipate the need for more than 65,536 code points (who would ever need more?); the main perceived threat to that limit having been resolved by Han unification.
Ok, but that doesn't answer the question; it's more of an indication that those design(at)s didn't uncover "the true extent" until years later